Re: [RFC PATCH 2/3] media: qcom: camss: Add CAMSS Offline Processing Engine driver

[Konrad Dybcio]

On 3/31/26 12:51 AM, Bryan O'Donoghue wrote:
> On 30/03/2026 19:55, Dmitry Baryshkov wrote:
>> This sounds like the most horrible idea: letting userspace directly
>> program any registers in a way that is not visible to the kernel.
> 
> No I'm wondering if there is a way to construct the basic format in user-space so it doesn't need to be re-interpreted stuffed/unstuffed.
> 
> As mentioned I believe there is a defunct qcom project which did/does just that, not sure why that hasn't been investigated/developed.

I believe this isn't a great idea since the format will at one point be
platform-dependent (I think it may be already) and one will have to teach
_all_ of the userspace implementations about all of these specifics

Unless I'm missing the bigger picture, we're not talking about super
large amounts of data that would need to be slightly shuffled around

Konrad